Re: Mersenne: SMT
Gareth Randell wrote: Some SMT news that I know of: Alpha EV8 will have SMT with 4 simultaneous execution paths. Alpha recently got canned by compaq, so the above may never happen. . and Intel bought Alpha from Compaq recently. Yours, Dieter Schmitt _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: SV: Mersenne: number of processors participating
On Tue, 30 Oct 2001 21:33:59 -0500, Rick Pali [EMAIL PROTECTED] wrote: Aaron Blosser wrote: Good old sysinternals... they have the neatest tools. Damn straight! I've been using (and loving) PageDefrag since I stumbled on that site. A few other gems have since made their way onto my system... Rick. See also HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management\ClearPageFileAtShutdown This setting was put in place by a 'twink' program I tried a month or two back, and has been working great for me (marginally more secure too). Nathan _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: mprime for Solaris x86 ?
George Woltman wrote: Just grab the source at http://www.mersenne.org/source.htm and see if it compiles and links. You will not compile the ASM stuff - both ELF and COFF style object files are provided for the ASM code. Er, my Solaris x86 machine just expired (suspect failed PSU), so maybe someone else can try this! :-( -- === Gareth Randall === _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: SMT
On 3 Nov 2001, at 21:40, Kel Utendorf wrote: At 21:01 11/03/2001 -0500, George Woltman wrote: Can prime95 take advantage of SMT? I'm skeptical. If the FFT is broken up to run in two threads, I'm afraid L2 cache pollution will negate any advantage of SMT. Of course, I'm just guessing - to test this theory out we should compare our throughput running 1 vs. 2 copies of prime95 on an SMT machine. I'm not sure I fully understand the way in which a SMT processor would utilise cache. But I can't see how the problem could be worse than running two copies of a program on a SMP system. This seems to work fairly well in both Windows and linux regimes (attatching a thread to a processor and therefore its associated cache, rigidly in the case of Windows, loosely but intelligently in the case of linux). If an SMT processor has a unified cache, cache pollution should surely be not too much of a problem? Running one copy thereby getting benefit of the full cache size would run that one copy faster, (just as happens with SMP systems where memory bandwidth can be crucial) but the total throughput with two copies running would surely be greater. Especially on a busy system, where two threads get twice as many timeslices as one! If there is some way in which the FFT could be broken down into roughly equal sized chunks, it _might_ be worth synchronizing two streams so that e.g. transform in on one thread was always in parallel with transform out on the other, and vice versa. Obviously you'd need to be running on two different exponents but using the same FFT length to gain from this technique. Whether this would be any better than running unsynchronized would probably require experimentation. Could things be setup so that factoring and LL-testing went on simultaneously? This would speed up the overall amount of work being done. Because trial factoring, or P-1/ECM on _small_ exponents, have a very low memory bus loading, running a LL test and factoring in parallel on a dual-processor SMP system makes a lot of sense. I suspect the same situation would apply in an SMT environment. The problem of mass deployment (almost everyone in this position, instead of only a few of us) is that there is a great deal of LL testing effort required in comparison to trial factoring, so running two LL tests in parallel but inefficiently would bring us to milestones faster than the efficient LL/trial factoring split. Regards Brian Beesley _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
SV: SV: Mersenne: Strange factor arrived though not calculating it
Yes I did check all prime.log and all results.txt. I would never have missed it if it had been mentioned in any of these files. You probably also did not notice that I have all results I have produced going to a special file called: Results.all (how creative I am :-) ), and this process is done automagicly by a script? br tsc OK. Did you look through the contents of prime.log on whichever system once had exponent 16871993? If the exponent was dropped because PrimeNet informed the system the job was already done, the evidence will be in there. If the system really did run that exponent, found the factor informed PrimeNet, the evidence will also be in there. Regards Brian Beesley _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
RE: Mersenne: SMT
I kind of like the practice that many dual processor folks have seem to adopted (and one which I'll be switching my group of computers too)... Namely, on dual CPU systems, have one Prime process doing LL tests, and have the other one doing trial factoring. Even on Compaq servers that have GREAT cache/memory management, running 2 LL tests on each CPU will slow down both processes. Running one LL and one factor reduces the hit on the memory subsystem since the factoring can generally remain in the CPU cache of it's respective processor, leaving the LL process to better use the memory for itself. So perhaps this same approach could be adopted for SMT? And just a reminder... trial factoring is still a great use of slower machines... I have an AMD K6-III 400 that can trial factor the current 17M exponents in just about 2 days. Yeah, P4's can do it a lot faster, or 1.2GHz Athlons, but I'd rather have those machines concentrate on the LL tests. Aaron -Original Message- From: [EMAIL PROTECTED] [mailto:mersenne-invalid- [EMAIL PROTECTED]] On Behalf Of [EMAIL PROTECTED] Sent: Sunday, November 04, 2001 1:36 PM To: Kel Utendorf Cc: [EMAIL PROTECTED] Subject: Re: Mersenne: SMT On 3 Nov 2001, at 21:40, Kel Utendorf wrote: At 21:01 11/03/2001 -0500, George Woltman wrote: Can prime95 take advantage of SMT? I'm skeptical. If the FFT is broken up to run in two threads, I'm afraid L2 cache pollution will negate any advantage of SMT. Of course, I'm just guessing - to test this theory out we should compare our throughput running 1 vs. 2 copies of prime95 on an SMT machine. I'm not sure I fully understand the way in which a SMT processor would utilise cache. But I can't see how the problem could be worse than running two copies of a program on a SMP system. This seems to work fairly well in both Windows and linux regimes (attatching a thread to a processor and therefore its associated cache, rigidly in the case of Windows, loosely but intelligently in the case of linux). If an SMT processor has a unified cache, cache pollution should surely be not too much of a problem? Running one copy thereby getting benefit of the full cache size would run that one copy faster, (just as happens with SMP systems where memory bandwidth can be crucial) but the total throughput with two copies running would surely be greater. Especially on a busy system, where two threads get twice as many timeslices as one! If there is some way in which the FFT could be broken down into roughly equal sized chunks, it _might_ be worth synchronizing two streams so that e.g. transform in on one thread was always in parallel with transform out on the other, and vice versa. Obviously you'd need to be running on two different exponents but using the same FFT length to gain from this technique. Whether this would be any better than running unsynchronized would probably require experimentation. Could things be setup so that factoring and LL-testing went on simultaneously? This would speed up the overall amount of work being done. Because trial factoring, or P-1/ECM on _small_ exponents, have a very low memory bus loading, running a LL test and factoring in parallel on a dual-processor SMP system makes a lot of sense. I suspect the same situation would apply in an SMT environment. The problem of mass deployment (almost everyone in this position, instead of only a few of us) is that there is a great deal of LL testing effort required in comparison to trial factoring, so running two LL tests in parallel but inefficiently would bring us to milestones faster than the efficient LL/trial factoring split. Regards Brian Beesley _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime?
What will we do when anyone finds a number of 10 million+ digits which is prime? Will everybody just leave the project because there is no prize to gain any longer? After the introduction of search for 10 million digits number this could leave the project with quite a big hole, say from M14.xxx.xxx and up till the exponent found. It will be kind of difficult to find new volunteers that will use time and electricity to fill the hole if nothing more than glory is won. Will there be an other prize? Will there be a new goal? We have the 4 or 5 biggest primes, are anyone stressing us by using another algorithm like the LL, have access to more CPU, working on a bulletproof method of generating new primes? br tsc _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
On Mon, Nov 05, 2001 at 12:02:30AM +0100, Torben Schlüntz wrote: What will we do when anyone finds a number of 10 million+ digits which is prime? Spread the prize according to the rules on the website, and continue searching? ;-) Will everybody just leave the project because there is no prize to gain any longer? Considering that the project worked very well for years (actually, a whole lot longer than we've had a prize) before EFF announced its prize, I doubt so. :-) After the introduction of search for 10 million digits number this could leave the project with quite a big hole, say from M14.xxx.xxx and up till the exponent found. Well, then we'll do our best to fill such a hole, won't we? :-) It will be kind of difficult to find new volunteers that will use time and electricity to fill the hole if nothing more than glory is won. Doubtful (see my comments above). Will there be an other prize? Will there be a new goal? EFF has a prize for a 100+ million digit prime, and a 1000+ million digit prime. The first prize (which has already been awarded) also went out to a GIMPS member, discovering the first 1+ million digit prime. :-) /* Steinar */ -- Homepage: http://www.sesse.net/ _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime?
At 12:02 AM 11/5/2001 +0100, =?utf-8?Q?Torben_Schl=C3=BCntz?= wrote: It will be kind of difficult to find new volunteers that will use time and electricity to fill the hole if nothing more than glory is won. I'm not in it for the prize, and a lot of others aren't either. I've been it over 4 years (maybe over 5 years?), well before there were any prizes. I just replaced a computer that I have been using over 4 years, and I think I was in GIMPS before I had that one. +-+ | Jud McCranie| | | | Programming Achieved with Structure, Clarity, And Logic | +-+ _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
RE: Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime?
Well, there's still plenty of people who'll stick to it. There's a lot of hardcore GIMPSers who are more than willing to keep crunching numbers alive (some even willing to risk getting the FBI on their case) :) Besides the big ticket things like finding the megadigit prime and a ten megadigit prime, there's still all the other work of testing all the exponents in between, confirming that Mersenne Prime XX is really the XXth prime (and we didn't just miss any in-between). There's the double-checking work, and I still think it's cool when I find a really big factor (anything over 80 bits is fun to find) of a Mersenne number. I'm sure a lot of folks would just give up, but personally, I'd contribute to some fund for a prize for the other Prime numbers we find along the way. Just my 2 cents worth. -Original Message- From: [EMAIL PROTECTED] [mailto:mersenne-invalid- [EMAIL PROTECTED]] On Behalf Of Torben Schlüntz Sent: Sunday, November 04, 2001 3:03 PM To: [EMAIL PROTECTED] Subject: Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime? What will we do when anyone finds a number of 10 million+ digits which is prime? Will everybody just leave the project because there is no prize to gain any longer? After the introduction of search for 10 million digits number this could leave the project with quite a big hole, say from M14.xxx.xxx and up till the exponent found. It will be kind of difficult to find new volunteers that will use time and electricity to fill the hole if nothing more than glory is won. Will there be an other prize? Will there be a new goal? We have the 4 or 5 biggest primes, are anyone stressing us by using another algorithm like the LL, have access to more CPU, working on a bulletproof method of generating new primes? br tsc _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: SMT
At 09:35 PM 11/4/2001 +, [EMAIL PROTECTED] wrote: I'm not sure I fully understand the way in which a SMT processor would utilise cache. This is a prime95 problem, not a SMT problem. Prime95 is designed to run efficiently in 128KB of L2 cache. If I split the current FFT into 2 threads, then either each thread is going to want 128KB of L2 cache space (more cache contention) or I must recode the passes of prime95 to run efficiently in just 64KB of cache (this might be a bad idea as using less L2 cache may require adding another pass over the FFT data - at least for some FFT sizes). That is why I fear that SMT may not be helpful to prime95. Another way of saying this is prime95 may be more constrained by L2 cache sizes than it constrained by micro-op scheduling. However, I had not considered the idea of factoring and LL testing. Even better, if we make the factoring code use mostly integer instructions then the LL test thread would keep the FPU units busy and the factoring thread would keep the integer units busy! Then the only question becomes Does the user want to slow down his LL tests in order to increase his total (LL + factoring) throughput. -- George _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime?
I'm sure a lot of folks would just give up, but personally, I'd contribute to some fund for a prize for the other Prime numbers we find along the way. Just my 2 cents worth. If we put 2 cents for every computer in GIMPS we would get $600 total :) I would be thrilled to get $600 for finding a prime... /Lars _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: What will we do when anyone finds a number of 10 million+ digits which is prime?
Hi, At 12:02 AM 11/5/2001 +0100, =?utf-8?Q?Torben_Schl=C3=BCntz?= wrote: What will we do when anyone finds a number of 10 million+ digits which is prime? Will everybody just leave the project because there is no prize to gain any longer? Careful reading of the prize rules show that $20,000 would go to GIMPS primarily to fund future awards. My goal would be $5000 per prime. That should keep us funded for a decade. Of course this is all very wishful thinking. It would take about 20,000 top-of-the-line P4 systems a year to have a 50% chance of finding a 10M prime. We will find it one day, but it is more likely to be when 5 and 10 GHz P4s are commonplace. _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
On Sun, Nov 04, 2001 at 07:27:35PM -0500, George Woltman wrote: Of course this is all very wishful thinking. It would take about 20,000 top-of-the-line P4 systems a year to have a 50% chance of finding a 10M prime. We will find it one day, but it is more likely to be when 5 and 10 GHz P4s are commonplace. Speaking of which -- shouldn't we be (statistically) really close to finding a new prime soon? /* Steinar */ -- Homepage: http://www.sesse.net/ _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
At 01:43 AM 11/5/2001 +0100, Steinar H. Gunderson wrote: Speaking of which -- shouldn't we be (statistically) really close to finding a new prime soon? Yes, statistically. You'd expect the next one to be before 14,000,000 and I've got assignments in the 13,000,000 range. However, all exponents have been checked once only to a little past 8,000,000. +-+ | Jud McCranie| | | | Programming Achieved with Structure, Clarity, And Logic | +-+ _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
On Sun, 04 Nov 2001 20:09:20 -0500, Jud McCranie [EMAIL PROTECTED] wrote: At 01:43 AM 11/5/2001 +0100, Steinar H. Gunderson wrote: Speaking of which -- shouldn't we be (statistically) really close to finding a new prime soon? Yes, statistically. You'd expect the next one to be before 14,000,000 and I've got assignments in the 13,000,000 range. However, all exponents have been checked once only to a little past 8,000,000. Of course, this whole argument makes (as far as I can see) heavy use of the gamblers' fallacy, aka the fallacy of maturation of probabilities (Hey, I lost the last 50 games - what are the odds against me losing 51 5-man games in a row? I'm certain to win!) Nathan _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
RE: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
Hmm... in games of chance, each game's outcome is independent of the results of past games, so that is a valid point. In Mersenne Prime hunting though, I think it's safe to say that statistically there should be some prime numbers in the range we're checking, so the more we *don't* find, the higher the odds of the remaining ones being prime. :) In other words, there is a certain sort of dependence on previous outcomes. In gambling terms, that's like saying that perhaps you could guarantee winning one game out of every 50 hands of poker (alright, so he's a lousy player). If you lost 49 times, and you know you would win 1 out of 50 times, then yes, there's a 100% chance you'll win the next hand. :) What's a bigger issue here is whether or not the statistical model for how many primes we expect to find in a given range is accurate or not. Since the probabilities are based purely on the spread of previous primes, the sample data is pretty small, and there's already a jagged curve to the whole thing, so any probabilities are likely to be off by a good amount in practice. I know we did lots of analyses prior to finding the last one, and I'm curious how well that # fit any of the odds people had formulated. Aaron -Original Message- From: [EMAIL PROTECTED] [mailto:mersenne-invalid- [EMAIL PROTECTED]] On Behalf Of Nathan Russell Sent: Sunday, November 04, 2001 5:49 PM To: [EMAIL PROTECTED] Subject: Re: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime? On Sun, 04 Nov 2001 20:09:20 -0500, Jud McCranie [EMAIL PROTECTED] wrote: At 01:43 AM 11/5/2001 +0100, Steinar H. Gunderson wrote: Speaking of which -- shouldn't we be (statistically) really close to finding a new prime soon? Yes, statistically. You'd expect the next one to be before 14,000,000 and I've got assignments in the 13,000,000 range. However, all exponents have been checked once only to a little past 8,000,000. Of course, this whole argument makes (as far as I can see) heavy use of the gamblers' fallacy, aka the fallacy of maturation of probabilities (Hey, I lost the last 50 games - what are the odds against me losing 51 5-man games in a row? I'm certain to win!) Nathan _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
At 08:48 PM 11/4/2001 -0500, Nathan Russell wrote: Of course, this whole argument makes (as far as I can see) heavy use of the gamblers' fallacy, aka the fallacy of maturation of probabilities (Hey, I lost the last 50 games - what are the odds against me losing 51 5-man games in a row? I'm certain to win!) Statistically, each exponent resulting in a prime is about twice the previous one. There is a heuristic argument that supports that being the case. The last one was 6.9 million something, we're now testing close to twice that. I think there are only three cases where one exponent is more than twice as large as the previous one. +-+ | Jud McCranie| | | | Programming Achieved with Structure, Clarity, And Logic | +-+ _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
Re: Mersenne: Re: What will we do when anyone finds a number of 10 million+ digits which is prime?
On Sun, 04 Nov 2001 21:46:58 -0500, Jud McCranie [EMAIL PROTECTED] wrote: At 08:48 PM 11/4/2001 -0500, Nathan Russell wrote: Of course, this whole argument makes (as far as I can see) heavy use of the gamblers' fallacy, aka the fallacy of maturation of probabilities (Hey, I lost the last 50 games - what are the odds against me losing 51 5-man games in a row? I'm certain to win!) Statistically, each exponent resulting in a prime is about twice the previous one. There is a heuristic argument that supports that being the case. The last one was 6.9 million something, we're now testing close to twice that. I think there are only three cases where one exponent is more than twice as large as the previous one. P1 P2 ratio 127 521 4.102362205 607 12792.10708402 442396892.190594619 216091 756839 3.502408707 1398269 2976221 2.128503886 and, almost certainly, 3021377 6972593 2.307753385 It looks like there's some clumpiness around ratios of 2.1, but I suspect that's an artifact of the cutoff point you gave. I wouldn't know how to check statistically, and the sample size is probably too small anyhow. I don't know whether the second-largest gap is an outlier, but the largest definately is. It might be noted that there are only 98-31-1=66 primes between 127 and 521 (exclusive); I don't know offhand whether this is significant. The gaps with ratio 1.1 follow, for those who are curious. 442342531.039971785 994196891.026008876 21701 19937 1.088478708 23209 21701 1.069489885 3021377 2976221 1.01517226 I don't see any outliers here. I might note that gaps of this size aren't even possible until after 29 (31/29=1.077, and 19/17=1.118, thus excluding gaps beginning with the first seven Mersennes before the primes above them are even calculated) Any comments from the list? Nathan _ Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers